13 February 2012

The Tumbi Quarry landslide – on the difference between trigger and cause

Posted by Dave Petley

There continues to be a moderate level of media interest in the Tumbi Quarry landslide in Papua New Guinea, with most of that focusing on the role of the quarryLNG Watch have examined some of the documents that the project operators themselves have online and have shown that audits of the project last year raised questions about the stewardship of the quarry site.  It is not entirely clear as to the nature of these issues, but the possibility that the quarry operations were inadequately overseen are clearly of interest here.

There does seem to be some confusion about the difference between the cause and the trigger in the case of this landslide.  It seems that there may be a line of argument developing that because the quarry was not in operation at the time of the landslide it cannot be responsible for it.  This is a fallacious case to make, if that is beig argued.  The issue here really does come down to the important distinction between the cause of a landslide and the trigger event that initiated it.  So let’s take a look at what we mean by these two concepts:

Trigger – the trigger event is the external or internal process that makes a slope to collapse at that particular point in time.  In most cases the trigger event is either heavy rainfall or an earthquake, although it can be human activity, snowmelt, etc.  We should note that some landslides have no external trigger, but in the majority of cases we can identify one.  As an analogy we might think of the demolition of a building – the trigger is someone pushing the button that fires the detonators.

Cause – the causes of a landslide are those processes that have made the slope susceptible to failure.  In the case of a natural slope these might be weak rocks, erosion at the toe by a raiver, deforestation, etc.  In most cases a combination of causes combine to make a slope unstable, and the trigger event  then initiates the failure.  Going back to the analogy of the demolition of a building, the causes are the placing of explosives, removal of structural walls, etc.

The key idea here is that both causes and the trigger combine to create the landslide event. Without the causes the triggercould not initiate the failure.  So in understanding a landslide event we need to identify not just the trigger, but to find out what caused the slope to be be ready to fail.  This is the role of an investigation.

It is also worth noting that in managing landslides, we often focus on both the causes and the trigger – for example we might install rock anchors to strengthen the soil (removing a cause) and drains to reduce pore pressure (blunting the effects of a trigger).

In the case of the Tumbi Quarry landslide, this distinction is important.  In investigating the landslide we must not consider just the trigger.  It may well be that the quarry was inactive at the time of the landslide, and that exceptionally heavy rainfall had occurred.  In this case the trigger might be the rainfall and not the quarry.  However, this does not give us information about the causes, and in the context of lives being lost it must be the causes that are important.  Here we need to look at the combination of other things that have been occurring to weaken the slope.  And in this context a key factor might be the role of the quarry.  It may well be that it in fact did not have anything to do with the final collapse, but we cannot say that without a proper forensic investigation.  Isolating the trigger does not provide the answers to these deeper questions.

For what it is worth, and from 12,000 km away, I continue to be a little perplexed by the explanation that this is a natural failure triggered by rainfall.  First, it just seems to be a remarkable coincidence that the only reported large-scale slope failure was a site that had been recently disturbed by human activity.  Coincidences do occur of course, but by definition they are rare, and scientists are rarely happy with an explanation that in effect says that it was an act of god.  Second, usually when a trigger event initiates a large, entirely natural failure failure like this it also triggers many, many smaller failures (as described by power law distributions).  Maybe the reported rainfall event did trigger many small failures, but as yet I have not seen any reports of this.  If it did not, then one would have even more reason to ask what happened on this particular slope to make it susceptible to failure.

I remain absolutely open-minded about the causes of this landslide, and continue to emphasise that it might be an entirely natural event.  A proper investigation will ascertain this without great difficulty.